基于改进多尺度Retinex的模糊图像增强仿真

Fuzzy Image Enhancement Simulation Based on Improved Multi-Scale Retinex

针对传统模糊图像增强方法未对图像进行降噪处理,导致增强后的图像出现视觉效果差、质量差、峰值信噪比低的问题,提出基于改进多尺度Retinex的模糊图像增强方法。结合阈值法对增强前的模糊图像进行降噪处理,提高图像质量;利用背景强度和梯度信息对降噪后的模糊图像进行二维分解,并采用改进多尺度Retinex对模糊图像进行照度分割,根据各子图的照度特性,采用不同的尺度的Retinex划分模糊图像各个区域,实现模糊图像增强。仿真结果表明,所提方法呈现的图像视觉效果佳、图像质量好、峰值信噪比高,具有一定的实际应用性。

The neglect of noise reduction in traditional fuzzy image enhancement methods leads to poor visual effect and quality, low signal-to-noise ratio. Therefore, this paper proposes a fuzzy image enhancement method based on improved multi-scale Retinex. Firstly, the threshold method was introduced to denoise the blurred image for improving the image quality. Secondly, according to the background intensity and gradient information, the denoised blurred image was decomposed into two dimensions. Then, the improved multi-scale Retinex was used to segment the blurred image. Finally, according to the illumination characteristics of each sub image, Retinex of different scales was applied to divide each region of the blurred image, thus completing the enhancement of the blurred image. The simulation results show that the method has excellent image visual effect and image quality, high peak signal-to-noise ratio and outstanding practical application.